90-Days mortality rate in patients treated within the context of a phase-I trial: how should we identify patients who should not go on trial?